OpenAI Launches GPT-5.4 Mini and Nano for Coding and Subagents
Smaller, faster models aim to bring GPT 5.4 capabilities to high volume API and agent workloads.
OpenAI on Tuesday released GPT-5.4 mini and nano, its most capable small models to date, designed for coding workflows and high-volume API deployments.
According to OpenAI, GPT-5.4 mini runs more than 2x faster than GPT-5 mini while approaching the performance of the full GPT-5.4 model on several coding benchmarks. On SWE-Bench Pro, which tests real-world software engineering tasks, GPT-5.4 mini scored 54.4% compared to 57.7% for the larger GPT-5.4.
GPT-5.4 nano targets the lowest-cost tier for classification, data extraction, and simpler coding subagents. In the API, GPT-5.4 mini costs $0.75 per million input tokens and $4.50 per million output tokens, while nano runs $0.20 and $1.25 respectively.
The models feature a 400,000-token context window and are available across OpenAI's API, Codex coding assistant, and ChatGPT (for free users via the Thinking feature).
The release reflects a broader industry shift toward model tiering—offering smaller, cheaper models optimized for latency-sensitive workloads rather than always reaching for the largest, most expensive model.
"GPT-5.4 mini delivers strong end-to-end performance for a model in this class," said Aabhas Sharma, CTO at Hebbia, which tested the model. "It matched or exceeded competitive models on several output tasks and citation recall at a much lower cost."
Primary source: OpenAI Blog